University of Wollongong

Research Online
Faculty of Informatics - Papers (Archive) Faculty of Engineering and Information Sciences

2007

Modulation solutions for nematicon propagation in

non-local liquid crystals
Antonmaria Minzoni
Universidad Nacional Autónoma de México

Noel Smyth
University of Edinburgh

Annette L. Worthy
University of Wollongong, annie@uow.edu.au

Publication Details
Minzoni, A., Smyth, N. F. & Worthy, A. L. (2007). Modulation solutions for nematicon propagation in non-local liquid crystals.
Optical Society of America. Journal B: Optical Physics, 24 (7), 1549-1556.

Research Online is the open access institutional repository for the University of Wollongong. For further information contact the UOW Library:
research-pubs@uow.edu.au
Modulation solutions for nematicon propagation in non-local liquid
crystals
Abstract
The propagation of solitary waves, so-called nematicons, in a nonlinear nematic liquid crystal is considered in
the nonlocal regime. Approximate modulation equations governing the evolution of input beams into steady
nematicons are derived by using suitable trial functions in a Lagrangian formulation of the equations for a
nematic liquid crystal. The variational equations are then extended to include the effect of diffractive loss as
the beam evolves. It is found that the nonlocal nature of the interaction between the light and the nematic has
a significant effect on the form of this diffractive radiation. Furthermore, it is this shed radiation that allows the
input beam to evolve to a steady nematicon. Finally, excellent agreement is found between solutions of the
modulation equations and numerical solutions of the nematic liquid-crystal equations.

Disciplines
Physical Sciences and Mathematics

Publication Details
Minzoni, A., Smyth, N. F. & Worthy, A. L. (2007). Modulation solutions for nematicon propagation in non-
local liquid crystals. Optical Society of America. Journal B: Optical Physics, 24 (7), 1549-1556.

This journal article is available at Research Online: http://ro.uow.edu.au/infopapers/2653

Minzoni et al. Vol. 24, No. 7 / July 2007 / J. Opt. Soc. Am. B 1549

Modulation solutions for nematicon propagation

in nonlocal liquid crystals

Antonmaria A. Minzoni,1 Noel F. Smyth,2,* and Annette L. Worthy3

1
Fenomenos Nonlineales y Mecánica, Department of Mathematics and Mechanics, Instituto de Investigación en
Matemáticas Aplicadas y Sistemas, Universidad Nacional Autónoma de México, Apdo. 20-726, 01000 México, D.F.
2
School of Mathematics and Maxwell Institute for Mathematical Sciences, The King’s Buildings, University of
Edinburgh, Edinburgh, Scotland, UK, EH9 3JZ
3
School of Mathematics and Applied Statistics, University of Wollongong, Northfields Avenue, Wollongong, New
South Wales, Australia, 2522
*Corresponding author: N.Smyth@ed.ac.uk

Received December 21, 2006; revised February 26, 2007; accepted February 27, 2007;
posted March 15, 2007 (Doc. ID 78332); published June 15, 2007
The propagation of solitary waves, so-called nematicons, in a nonlinear nematic liquid crystal is considered in
the nonlocal regime. Approximate modulation equations governing the evolution of input beams into steady
nematicons are derived by using suitable trial functions in a Lagrangian formulation of the equations for a
nematic liquid crystal. The variational equations are then extended to include the effect of diffractive loss as
the beam evolves. It is found that the nonlocal nature of the interaction between the light and the nematic has
a significant effect on the form of this diffractive radiation. Furthermore, it is this shed radiation that allows
the input beam to evolve to a steady nematicon. Finally, excellent agreement is found between solutions of the
modulation equations and numerical solutions of the nematic liquid-crystal equations. © 2007 Optical Society
of America
OCIS codes: 190.5530, 190.5940, 160.3710, 190.4400.

1. INTRODUCTION
The possibility of guiding self-supporting, nonlinear
beams (solitary waves) of light in the regime of large non-
locality in liquid crystals was shown in a remarkable se-
ries of experiments by Assanto et al. [1–4]. They named
these light solitons, which resulted from the liquid crystal
actively guiding the light, nematicons. Moreover, they
also studied nematicon formation theoretically [2,3]. In
this work the existence of static solitons was shown by us-
ing an expansion around the center of the beam, which
gave a Gaussian beam and a parabolic profile for the op-
tical axis [2,3]. This local expansion was matched to an
exponentially decaying solution away from the center of
the beam. This matching then gave a condition determin-
ing the relation between the nematicon’s amplitude and
the width of its waist. It also showed that the angle (di-
rector) profile is much wider than the waist of the beam.
The dynamics of a nematicon was studied approximately
by using the Fock–Leontovich equation in the limit of a
large cell [3]. An expression for the evolution of the waist
of the beam was then obtained, which showed its oscilla-
tion. It was further shown that steady solitons with no
waist oscillation are recovered for specific values of the in-
put power. The effect of diffractive loss by the soliton as it
evolves was not considered.
On the more abstract side, it is known, since the nem-
aticon equations (1) and (2) below are the same as those
for a thermoelastic waveguide, that the soliton solution is
stable [5]. However, no results for the dynamics of the
evolution of a beam-type boundary condition can be ob-
tained using the Lyapunov function discussed in Ref. [5].
The purpose of the present work is to fill the gap be-
tween these abstract and approximate results. To this
end, the dynamics of nematicon evolution in a liquid crys-
tal is studied using an extension of the method of Ref. [6]
for optical pulse propagation, which is based on a form of
modulation theory that includes the effect of the disper-
sive radiation shed by a pulse solitary wave during its
evolution. This method was further developed to apply to
soliton propagation in liquid crystals in the local regime
[7,8]. In a similar work, but for one rather than for two
space dimensions, as in Refs. [7] and [8], a variational
method was employed to obtain approximate Gaussian
profiles for steady one-dimensional nematicons [9].
Gaussian profiles were found to be a good approximation
near the peak of a one-dimensional nematicon but de-
cayed too fast in its tail. Lagrangian methods were also
employed to find approximations to the steady two-
dimensional solitary wave solutions of a nonlocal nonlin-
ear Schrödinger (NLS) equation with a Gaussian kernel
in the nonlinear, nonlocal term [10]. However, such a
Gaussian kernel is not applicable to liquid crystals.
An initial beam evolves into a nematicon by undergoing
a breathing motion accompanied by the shedding of linear
diffractive radiation. The modulation equations for the
beam evolution are obtained by using a suitable trial
function in an averaged Lagrangian for the nematicon
equations. These modulation equations have a family of
fixed points that are the steady nematicon solutions, with
conservation of energy selecting the final steady state. It
is shown how the modulation equations, coupled with loss
due to shed diffractive radiation, explain and reproduce

0740-3224/07/071549-8/$15.00 © 2007 Optical Society of America

1550 J. Opt. Soc. Am. B / Vol. 24, No. 7 / July 2007
Fig. 1. (Color online) Schematic diagram of a liquid-crystal cell
with a propagating polarized light beam.
quantitatively numerical solutions of the full equations
governing the evolution of a nematicon.
2. FORMULATION
Let us consider a polarized beam propagating in the z di-
rection in a cell containing a nematic liquid crystal, such
as was studied in works by Assanto et al. [1–3] and by
García Reimbert et al. [7,8]. The geometry of the liquid-
crystal cell is the same as that described in Ref. [7] and is
shown in Fig. 1. The nematicon propagates in the z direc-
tion, and the light is polarized such that its electric field is
in the x direction. The optical axis of the nematic is
pretilted by a static electric field, also in the x direction,
with an angle ␪ˆ in order to overcome the Freédericksz
threshold. The deviation from the pretilt angle due to the
optical field is denoted by ␪. The slowly varying envelope
of the optical electric field is denoted by E. The equations
governing the propagation of a nematicon in the liquid-
crystal cell are then [2,7]
⳵E 1
i + ⵜ2E + sin共2␪兲E = 0, 共1兲
⳵z 2
␯ⵜ2␪ − q sin共2␪兲 = − 2兩E兩2 cos共2␪兲. 共2兲
Here q is related to the external static electric field, and ␯
is the normalized elastic coefficient [8].
In the nonlocal limit it is known that the spatial extent
of the response of the optical director is larger than the
waist of the beam [2,3]. Moreover, it was also found in this
same work that the deviation ␪ of the optical director
from the equilibrium value is small. Under these nonlocal
and small deviation assumptions, the nematicon equa-
tions (1) and (2) can be simplified to
⳵E 1
i + ⵜ2E + 2␪E = 0, ␯ⵜ2␪ − 2q␪ = − 2兩E兩2 , 共3兲
⳵z 2
which are the same equations as for the thermoelastic
waveguide [5]. As the parameter ␯ increases, the response
of the liquid crystal becomes more nonlocal, and the width
of the optical axis becomes wider than the waist of the
light beam. Note that in the local limit ␯ → 0, the system
(3) can be reduced to the two-dimensional NLS equation,
which predicts collapse of an input beam at a finite propa-
Minzoni et al.
gation distance z. The collapse of the beam in this local
limit is stopped by the saturating nonlinearity of the non-
linear nematicon equations (1) and (2) [8]. In the present
work, the dynamics of how nonlocality prevents collapse
for the nonsaturable nematicon equations (3) will be stud-
ied.
3. APPROXIMATE EQUATIONS
To derive the approximate equations describing the evo-
lution of a nematicon, the Lagrangian for the nematicon
equations (3) is taken in polar coordinates:
冕冕
Z ⬁
L= 关ir共E*Ez − EEz*兲 − r兩Er兩2 + 4r␪兩E兩2 − ␯r␪2r
0 0
− 2qr␪2兴dr dz, 共4兲
where the superscript *
denotes the complex conjugate.
This Lagrangian, which will be used to construct the ap-
proximate equations for nematicon evolution, is the same
as that for thermoelastic waveguides [5].
As in Refs. [6–8], trial functions for the electric field E
and director angle ␪ are taken in the form
E = a sech共r/w兲ei␴ + ig ei␴, ␪ = ␣ sech2共r/␤兲. 共5兲
The electric field amplitude a, width w, phase ␴, and the
director angle amplitude ␣ and width ␤ are functions of z.
The parameter g is also a function of z. The term
ig exp共i␴兲 in the trial function for the electric field E in
Eqs. (5) represents the shed low-wavenumber diffractive
radiation that sits under the evolving beam, forming a
flat shelf or pedestal. This low-wavenumber, linear dif-
fractive radiation has low group velocity and so cannot es-
cape from the vicinity of the beam [6–8]. Perturbed in-
verse scattering for the NLS equation shows that for an
optical pulse this low-wavenumber radiation is out of
phase with the pulse, hence the factor i in this radiation
term [6]. As the shed radiation cannot remain flat away
from the beam or pulse, it is assumed that g is nonzero in
a circular region of radius l centered at the beam [6–8].
The electric field E and director angle ␪ have half-
widths w and ␤, respectively, in the trial functions Eqs.
(5). In the strong nonlocal limit under consideration, it is
expected that ␤ Ⰷ w, which leads to a more complicated
system of modulation equations than for the local limit,
for which both the electric field and the director angle
have the same half-widths [8].
An averaged Lagrangian is now obtained by substitut-
ing the trial functions [Eqs. (5)] into the Lagrangian (4)
and evaluating the r integral. All the resulting integrals
can be simply evaluated, except for the cross integral:
冕 冕
⬁ ⬁
r r
4r␪兩E兩2dr = 4␣a2 r sech2 sech2 dr. 共6兲
0 0
␤ w
This integral cannot be evaluated exactly unless ␤ = w.
To obtain useful, explicit approximate equations, the con-
cept of equivalent functions is now used [8]. The func-
tion sech2共r / ␤兲 is then replaced by the Gaussian
exp共−r2 / 共A␤兲2兲, and sech2共r / w兲 is replaced by the Gauss-
ian exp共−r2 / 共Bw兲2兲, so that the integral (6) can now be ex-
plicitly evaluated. With this replacement, we have
Minzoni et al. Vol. 24, No. 7 / July 2007 / J. Opt. Soc. Am. B 1551

冕 冕 冕
⬁ ⬁ ⬁
2共1/共A␤兲2+1/共Bw兲2兲 2 1
4r␪兩E兩2dr = 4␣a2 r e−r dr I42 = x sech4 x tanh2 x dx = ln 2 + , 共12兲
0 0 0
15 60

2 ␣ a 2A 2B 2w 2␤ 2

冕
⬁
= . 共7兲 2 1
A 2␤ 2 + B 2w 2 I4 = x sech4 x dx = ln 2 − ,
0
3 6
The new coefficients A and B are now determined by
matching the Taylor series of the integral (6) with the hy- where C is the Catalan constant C = 0.915965594. . . [11].
perbolic secants to the result [Eq. (7)] using Gaussians to Taking variations of the averaged Lagrangian (10) with
two terms in the limit w Ⰶ ␤, which is the limit of interest respect to the parameters a, w, ␣, ␤, ␴, and g results in
in the nonlocal regime. This results in the variational equations

d
A = 2I2/冑Ix32, B = 冑2I2 , 共8兲 共I2a2w2 + ⌳g2兲 = 0, 共13兲
dz
where
d␴

冕
⬁
d
2
共I1aw2兲 = ⌳g , 共14兲
I2 = x sech x dx = ln 2, dz dz
0

dg I22a A2B4␣aw2␤2

冕
⬁ I1 = − , 共15兲
Ix32 = x3 sech2 xdx = 1.352301002 . . . . 共9兲 dz 2w2 共A2␤2 + B2w2兲2
0
d␴ I22 A2B2␣␤2共A2␤2 + 2B2w2兲
The error in this Taylor-series matching is O共共w / ␤兲3兲. For I2 =− + , 共16兲
the comparison example of Fig. 3, the steady-state values dz w2 共A2␤2 + B2w2兲2
of w and ␤ are w0 = 1.6768 and ␤0 = 3.9647, giving an error plus the algebraic equations
in matching of O共8 ⫻ 10−2兲. In general, the larger ␯ is, the
smaller 共w / ␤兲3 is. A 2B 2␤ 2w 2a 2
It might be thought that, since the hyperbolic secants ␣= , 共17兲
2共A2␤2 + B2w2兲共2␯I42 + qI4␤2兲
of the trial functions have been replaced by equivalent
Gaussians to evaluate the cross integral, it might have
been better to use Gaussians for the trial functions [Eqs. A 2B 4w 4a 2
␣= .
(5)]. However, it was found that if this were done the re- qI4共A2␤2 + B2w2兲2
sulting approximate equations would not give solutions in
good agreement with numerical solutions, as was also the These variational equations are the equations governing
case for the nematicon equations in the local regime [8]. the evolution of the nematicon.
Furthermore, it was found that Gaussian initial condi- By Nöther’s theorem, the nematicon equations (3) have
tions have a greater tendency to split into multiple soli- an energy conservation equation that can be obtained
tons at moderate-to-high amplitudes than do hyperbolic from the Lagrangian (4) from invariances in z. The aver-
secant initial conditions. aged form of this energy conservation equation is then

冕
With the equivalent Gaussian approximation for the ⬁
dH d
cross integral, the averaged Lagrangian is = r关兩Er兩2 − 4␪兩E兩2 + ␯␪2r + 2q␪2兴dr
dz dz 0
L = − 2共a2w2I2 + ⌳g2兲␴⬘ − 2I1aw2g⬘ + 2I1gw2a⬘ + 4I1awgw⬘

− a2I22 − 4␯I42␣2 − 2qI4␣2␤2 +

2A2B2␣a2␤2w2
A ␤ +B w
2 2 2 2
. 共10兲
=
d
dz
冉 I22a2 + 4␯I42␣2 + 2qI4␣2␤2 −
2A2B2␣a2w2␤2
A 2␤ 2 + B 2w 2
冊 = 0.

共18兲
Here
Since energy is conserved, this energy conservation equa-
1 2 tion can be used to determine the final steady nematicon
⌳= 2
l 共11兲
state from the input beam.
is the area of the shelf under the beam, modulo 2␲. The When ␣ is eliminated between the algebraic equations
integrals I22, I1, I42, and I4 are given by (17), the director angle width ␤ can be found as

qI4B2w2 + 冑q2I42B4w4 + 16␯qI42I4A2B2w2

冕
⬁
1 1 ␤2 = . 共19兲
2 2
I22 = x sech x tanh x dx = ln 2 + , 2qA2I4
0
3 6
With this expression for ␤, the amplitude ␣ of the director

冕
⬁ is then given by the first of Eqs. (17). The amplitude and
I1 = x sech x dx = 2C, width of the optical axis are therefore given as algebraic
0 functions of the nematicon in the electric field.
1552 J. Opt. Soc. Am. B / Vol. 24, No. 7 / July 2007 Minzoni et al.

At the fixed point of the approximate evolution equa- ⳵E 1 ⳵ 2E 1 ⳵E

tions (13)–(16), g = 0. Hence Eq. (15) for g gives that the
steady amplitude â and width ŵ of the nematicon are re-
lated by
I22共A2␤ˆ 2 + B2ŵ2兲3共2␯I42 + qI4␤ˆ 2兲
â2 = , 共20兲
A4B6␤ˆ 4ŵ6
which determines the family of steady nematicons with â
as a function of ŵ. With this relation between â and ŵ,
the energy equation (18) then determines the fixed point
ŵ from the parameters of the input beam. However, this
energy equation is a transcendental equation in ŵ and
has to be solved numerically.
To complete the modulation equations, the radius ᐉ of
the shelf of low-wavenumber diffractive radiation under
the evolving nematicon needs to be determined. As in pre-
vious work on NLS-type equations [6–8], this is achieved
by linearizing the modulation equations (13)–(18) about
the fixed points â, ŵ, ␣ˆ , and ␤ˆ . The fixed point of the
modulation equations is then found to be a center whose
frequency is related to ⌳. As in Ref. [6], which was later
confirmed by Ref. [12], the frequency of this center is then
equated to the soliton frequency ␴ˆ ⬘ at the fixed point,
which results in ⌳, and hence ᐉ, in terms of the fixed-
point nematicon parameters. This determination of ᐉ,
while straightforward in principle, is involved owing to
the fact that five variables are involved in the lineariza-
tion. The details are then given in Appendix A, with the
final result for ⌳ being
i + + = 0. 共22兲
⳵z 2 ⳵r 2
2r ⳵r
This linearized equation was solved using Laplace
transforms [7], so the details will not be repeated here.
However, there is an important difference in the shelf of
diffractive radiation under the E beam, this being due to
the nonlocal interaction between the electric field and the
director. This difference can be seen in Fig. 2, which
shows the full numerical solution of the nematicon equa-
tions (1) and (2). It can be seen that the shelf under the E
beam extends well beyond it and has the form of a trun-
cated cone. This extended shelf is due to the nonlocal in-
teraction with the director as the director beam has a long
tail due to the large value of ␯. This in turn forces the
electric field to have a long extension. The shelf of diffrac-
tive radiation sitting under the electric field beam there-
fore has two components. The inner portion of this shelf of
radius ᐉ was considered in Section 3 and is due to the
resonant interaction of zero-wavenumber diffractive ra-
diation with the beam. The outer portion of the shelf is
due to the nonlocal forcing of the optical axis and has a
radius ␳, which is related to the half-width ␤1/2
= sech−1共1 / 冑2兲␤ of the optical disturbance of width ␤. This
division of the shelf into two components, an inner shelf

␴ˆ ⬘I12共ŵ2 + 2âŵ⍜兲
⌳=− , 共21兲
Q

with ⍜ and Q given in Appendix A.

The set of modulation equations governing the evolu-
tion of the nematicon is then Eqs. (14)–(16), the first of
Eqs. (17) and Eqs. (18) and (19), with the shelf radius de-
termined by Eq. (21). Note that the mass variational
equation (13) has been replaced by the energy equation
(18), which can be done, since the energy equation can be
obtained by a suitable combination of the variational
equations (13)–(16). As shown in Appendix A, these modu-
lation equations predict that the nematicon will oscillate
around the fixed point as it evolves. However, these modu-
lation equations are not complete, as the effect of the lin-
ear diffractive radiation shed as the beam evolves has not
been included. The determination of the effect of this shed
radiation is the subject of Section 4. When this radiation
is included, the beam stabilizes to the steady state by un-
dergoing oscillations in its waist width [3], in the manner
of a damped simple harmonic oscillator.

4. DIFFRACTIVE RADIATION LOSS

To obtain good agreement with numerical solutions, the
effect of the linear diffractive radiation shed by the beam
as it evolves must be included [6–8]. The shed radiation
has small amplitude relative to the beam and, thus being
governed by the linearized electric field equation of (3), is
Fig. 2. Numerical solution nematicon equations (1) and (2) at
z = 400 for the initial values a = 0.5, w = 4, with ␣ and ␤ deter-
mined by the first of Eqs. (17) and Eq. (19). The parameter values
are q = 2 and ␯ = 10. (a) Solution for 兩E兩: solid curve; solution for ␪:
dashed curve. (b) Solution for Re共E兲: solid curve; solution for ␪:
dashed curve.
Minzoni et al.
and an outer shelf, was also found by Ref. [12] for per-
turbed vector solitons of the coupled NLS equations de-
scribing birefringent optical fibers. These two components
of the shelf were shown to be due to two distinct types of
eigenfunction of the continuous spectrum for the equa-
tions describing the perturbation.
The radius ␳ of the outer portion of the shelf under the
electric field beam can be estimated from the numerical
solution shown in Fig. 2 as approximately ␳ = 15. Solu-
tions of the modulation equations give that ␤
= 2.18853. . . at z = 400 for the initial conditions of Fig. 2.
Hence an estimation for the radius of the outer portion of
the shelf is ␳ = 7␤1/2, so that
˜ = 1 ␳2 ,
⌳ 共23兲
2
冑2␲I1
冕 兵关
z ˜兲
␦=−
˜
32eR⌳ 0
1
4
˜兲
ln共共z − z⬘兲/⌳
1 ⳵ 2␪
R2 = ˜ g2兴.
关I2a2w2 − I2â2ŵ2 + ⌳ 共26兲
˜
⌳ ⳵r
For the local regime considered by Refs. [7] and [8], ⌳ ˜
= ⌳ as the optical axis and the electric field have the same
width, and the shelf under the electric field beam is the
usual NLS shelf [6]. The loss coefficient [Eq. (25)] differs
from that of García Reimbert et al. [7,8] by the replace-
˜ in the loss coefficient ␦ and R. This is be-
ment of ⌳ by ⌳
cause ⌳ in these terms comes from the total radius of the
shelf under the beam [7].
The final modulation equations describing the evolu-
tion of the nematicon are then Eqs. (14) and (16), the first
of Eqs. (17), (18)–(19), and (24). The solution of these
modulation equations will be compared with full numeri-
cal solutions of the nematicon equations (1) and (2) in Sec-
tion 5.
5. RESULTS
In this section, full numerical solutions of the nematicon
equations (1) and (2) will be compared with solutions of
the approximate equations (14) and (16), the first of Eqs.
(17), (18)–(19), and (24). The numerical method used to
solve the nematicon equations (1) and (2) is the same as
that of Ref. [8], so only a brief description of the method
will be given here. The electric field equation (1) was
solved using a pseudospectral method [13]. The main dif-
ference with the method described in Ref. [13] is that the
stepping in the z direction is performed in Fourier space
using a fourth-order Runge–Kutta method rather than in
physical space using a leap-frog scheme. The r derivatives
are calculated using fast Fourier transforms. To numeri-
cally solve the optical axis equation (2), it was rewritten
in the form
Vol. 24, No. 7 / July 2007 / J. Opt. Soc. Am. B 1553
in analogy with the relation between ᐉ and ⌳. Even
though these values of ␳ and ⌳ ˜ have been found for a spe-
cific initial condition, they will be found to give good
agreement with numerical solutions for all initial condi-
tions.
The diffractive radiation loss is now added to the modu-
lation equations (14)–(16), the first of Eqs. (17), and Eqs.
(18) and (19), which results in Eq. (15) being modified to
[7,8]
dg I22a A2B4␣aw2␤2
I1 = − − 2␦g, 共24兲
dz 2w2 共A2␤2 + B2w2兲2
where the loss coefficient ␦ is
␲R共z⬘兲ln共共z − z⬘兲/⌳ dz⬘
, 共25兲
兴 2
其
+ 3␲2/16
2
˜ 兲兴2/16 共z − z⬘兲
+ ␲2关ln共共z − z⬘兲/⌳
␯ ⳵␪
␯ + − 2q␪ = q sin共2␪兲 − 2q␪ − 2兩E兩2 cos共2␪兲.
2
r ⳵r
共27兲
The derivatives were calculated using standard second-
order finite differences, which gave a tridiagonal system.
Together with the boundary conditions ␪r = 0 at r = 0 and
␪ → 0 as r → ⬁, the resulting two-point boundary-value
problem was solved using a Picard iteration. The modula-
tion equations (14), (16), (18), and (24) were solved nu-
merically using the standard fourth-order Runge–Kutta
scheme.
The first comparison between the numerical solution
and the solution of the approximate equations is shown in
Fig. 3, for which the initial conditions are that a = 0.5 and
w = 4 at z = 0 with q = 2 and ␯ = 10. It can be seen that there
is excellent agreement in both the mean and the envelope
of the amplitude oscillation, with the approximate mean,
being the amplitude of the final steady nematicon, being
slightly higher than the numerical mean. The agreement
between the period of the solution of the modulation equa-
tions and that of the full numerical solution is excellent.
This agreement is remarkable, given that the amplitude
of the beam doubles over the initial amplitude during the
oscillations and that the frequency of the modulation
equations’ solution was obtained by linearizing about the
fixed point and then assuming that this result holds away
from the fixed point. In the derivation of the modulation
equations, it was assumed that for large ␯ the director re-
sponse ␪ is small. While the electric field amplitude a has
a large amplitude oscillation, numerical solutions of the
full nematicon equations (1) and (2) show that the direc-
tor angle has a small oscillation amplitude of ␣ = O共0.2兲.
The reason that the electric field response is so large is
that two-dimensional solitary waves for nonlocal NLS-
1554 J. Opt. Soc. Am. B / Vol. 24, No. 7 / July 2007
Fig. 3. Amplitude a for the electric field E as a function of z for
the initial conditions a = 0.5, w = 4, with ␣ and ␤ determined by
the first of Eqs. (17) and Eq. (19). The parameter values are q
= 2 and ␯ = 10. Numerical solution of nematicon equations (1) and
(2): solid curve; solution of modulation equations (14) and (16),
the first of Eqs. (17), and Eqs. (18), (19), and (24): dashed curve.
Fig. 4. Amplitude a for the electric field E as a function of z for
the initial conditions a = 0.6, w = 4, with ␣ and ␤ determined by
the first of Eqs. (17) and Eq. (19). The parameter values are q
= 2 and ␯ = 10. Numerical solution of nematicon equations (1) and
(2): solid curve; solution of modulation equations (14) and (16),
the first of Eqs. (17), and Eqs. (18), (19), and (24): dashed curve.
type equations are close to instability. The amplitude
grows to quite large values, becoming unstable, before
nonlocality stops the growth [14].
The nematicon equations (1) and (2) have been nondi-
mensionalized by the Rayleigh distance, so that the nem-
aticon evolution of Figs. 3–6 is shown over much longer
distances than are experimentally possible. This was
done so that the figures show a number of oscillations and
a significant degree of evolution to the steady state. An-
other point is that loss cannot be ignored over such long
evolution distances [15,16]. Now that the basic approxi-
mate equations for nematicon evolution have been deter-
mined, the addition of dissipative terms could form the
basis for further work.
Figure 4 shows a similar amplitude comparison for the
initial conditions a = 0.6 and w = 4 with q = 2 and ␯ = 10.
The envelope and mean of the approximate and numeri-
cal solutions are again in good agreement, with, again,
the mean of the approximate solution (i.e., the final
steady state) being slightly higher. For this initial condi-
Minzoni et al.
tion, the period of the approximate solution is slightly
shorter than the numerical period. In Section 5 the radia-
tive loss was fixed by determining the extended length of
the shelf under the electric field beam from a specific nu-
merical example, which was for the initial condition of
Fig. 3. It can be seen that the decay rate of the oscillations
of the solution of the approximate equations is still in
very good agreement with the numerical decay rate for
this same value of ⌳ ˜.
Figure 5 shows the same amplitude comparison as for
the two previous figures for the initial values a = 0.7 and
w = 4 with q = 2 and ␯ = 10. Again, the mean of the approxi-
mate oscillations is slightly higher than the numerical
mean, with the decay rate in good agreement with the nu-
merical value. However, the period of the oscillations of
the modulation solution is somewhat shorter than the nu-
merical period, with an additional phase difference. As
the amplitude of the beam nearly quadruples during the
oscillation, this is remarkable agreement, given that the
approximate period was determined by linearization
about the fixed point. This example is further justification
for choosing the value of ⌳ ˜ from a specific example. As the
modulation equations form a nonlinear oscillator, the dif-
ference in the amplitudes as given by the numerical and
approximate solutions is linked to this difference in the
period. These comments also explain the amplitude and
period differences between the numerical and approxi-
mate solutions of Fig. 4.
The final amplitude comparison in Fig. 6 is for the
lower initial amplitude a = 0.4 with the same initial width
w = 4 and q = 2, ␯ = 10. In contrast to the three previous ex-
amples, there is not good agreement between the numeri-
cal and the approximate amplitudes, while there is still
good agreement in the period. The reason for the ampli-
tude disagreement is that this initial condition is close to
the threshold amplitude at a = 0.32 for w = 4, q = 2, and ␯
= 10, below which the input beam decays into diffractive
radiation and does not form a nematicon, in which case
good agreement is not expected. The oscillation period
agreement is better than in Figs. 4 and 5 as the amplitude
Fig. 5. Amplitude a for the electric field E as a function of z for
the initial conditions a = 0.7, w = 4, with ␣ and ␤ determined by
the first of Eqs. (17) and Eq. (19). The parameter values are q
= 2 and ␯ = 10. Numerical solution of nematicon equations (1) and
(2): solid curve; solution of modulation equations (14) and (16),
the first of Eqs. (17), and Eqs. (18), (19), and (24): dashed curve.
Minzoni et al. Vol. 24, No. 7 / July 2007 / J. Opt. Soc. Am. B 1555

sians were used to replace the hyperbolic secant trial

functions in this integral. Using this replacement, an ex-
plicit averaged Lagrangian could be obtained. Finally, as
remarked above, the nonlocality introduced an additional
feature in the shelf of diffractive radiation under the
evolving nematicon. The analysis undertaken shows how
the results are independent of the particular nematicon
system of the present work. The present analysis could
then be useful in studying other soliton problems that in-
volve some type of nonlocal behavior.

APPENDIX A: SHELF RADIUS

The radius ᐉ of the shelf of diffractive radiation under the
evolving nematicon is determined by linearizing the
Fig. 6. Amplitude a for the electric field E as a function of z for modulation equations (13)–(18) about the fixed point with
the initial conditions a = 0.4, w = 4, with ␣ and ␤ determined by
the first of Eqs. (17) and Eq. (19). The parameter values are q a = â + a1, w = ŵ + w1, ␣ = ␣ˆ + ␣1,
= 2 and ␯ = 10. Numerical solution of nematicon equations (1) and
(2): solid curve; solution of modulation equations (14) and (16), ␤ = ␤ˆ + ␤1, g = g 1, ␴ = ␴ˆ + ␴1 , 共A1兲
the first of Eqs. (17), and Eqs. (18), (19), and (24): dashed curve.
where 兩a1兩 Ⰶ â, 兩w1兩 Ⰶ ŵ, 兩␣1兩 Ⰶ ␣ˆ , 兩␤1兩 Ⰶ ␤ˆ , 兩g1兩 Ⰶ 1, and 兩␴1兩
in this case only doubles during the oscillations, rather Ⰶ ␴ˆ . In a straightforward, but tedious, calculation, it is
than triples as in these previous figures. The modulation found on eliminating all the derivatives in terms of a1⬘ and
equations are then closer to their linearized form about then eliminating a1⬘ in terms of g1 that
the fixed point, from which the shelf radius ᐉ, which de-
termines the oscillation period of the modulation equa- d 2g 1 Q⌳␴ˆ ⬘
tions, was determined. 2
− g1 = 0, 共A2兲
dz I12共ŵ2 + 2âŵ⍜兲

where
6. CONCLUSIONS
The evolution of beamlike initial conditions into steady 2I22D2â − A2B2D␤ˆ 2ŵ2â共2␣ˆ + ⌫1â兲
⍜= ,
nematicons has been studied for the regime of strongly
A2B2␤ˆ ŵâ2共2A2␣ˆ ␤ˆ 3 + ⌫2D␤ˆ ŵ + 2B2⌫3␣ˆ ŵ3兲
nonlocal nematic liquid crystals. Modulation equations
for the evolution of the beam were derived by using suit-
able trial functions in a Lagrangian formulation of the I22
Q = Q1 − 关Q2 + ⍜Q3兴,
nematicon equations. When terms that include the effect 2A2B4␣ˆ ␤ˆ 2ŵ4
of diffractive loss as the beam evolves were added to these

冉 冊
modulation equations, excellent agreement was found be-
I22 2â
tween numerical solutions and solutions of the modula- Q1 = 1− ⍜ , Q2 = ␣ˆ 2␤ˆ 2ŵ2 + ⌫1␤ˆ 2ŵ2â,
2
tion equations. In the strongly nonlocal regime studied in 2ŵ ŵ
the present work, the width of the beam in the optical
axis is much larger than that of the electric field. This
Q3 = 2␣ˆ ␤ˆ 2ŵâ + ⌫2␤ˆ 2ŵ2â + 2⌫3␣ˆ ␤ˆ ŵ2â
larger optical axis width has the effect of increasing the
length of the shelf of linear diffractive radiation under the − 4D−1␣ˆ ␤ˆ 2ŵ2â共A2⌫3␤ˆ + B2ŵ兲, 共A3兲
electric field beam over that studied in previous works
[6–8]. This, in turn, leads to increased diffractive loss
from the nematicon as it evolves. 2A2B4ŵ4â 4A4B4␤ˆ ŵ3â2共␤ˆ − ⌫3ŵ兲
⌫1 = , ⌫2 = ,
The approach of the present work completes that of qI4D2 qI4D3
Conti et al. [3], since the effect of the diffractive radiation
shed as the beam evolves is taken into account. It has
B2ŵ共qI4␤ˆ 2 + 4␯I3兲
been shown that the shed radiation acts as a damping ⌫3 = , D = A2␤ˆ 2 + B2ŵ2 . 共A4兲
that settles the initial condition to the steady nematicon. qI4␤ˆ 共2A2␤ˆ 2 − B2ŵ2兲
The evolving beam undergoes a breathing motion in its
waist, in the manner of a damped simple harmonic oscil- As in Ref. [6], the frequency of the simple harmonic mo-
lator. tion equation (A2) is now matched to the steady nemati-
As final comments, some remarks are needed on the con oscillation frequency ␴ˆ ⬘, given by Eq. (16) with a, w,
nature of the approximations made to derive the evolu- ␤, and ␣ replaced by their fixed-point values. This results
tion equations for the nematicon. The trial functions used in the radius ᐉ of the shelf of low-wavenumber diffractive
for the beam electric field and the director angle include radiation under the beam being determined by
the effect of nonlocality, introducing an additional varia-
␴ˆ ⬘I12共ŵ2 + 2âŵ⍜兲
tional parameter. To evaluate a coupling integral between ⌳=− . 共A5兲
the electric field and the director angle, equivalent Gaus- Q
1556 J. Opt. Soc. Am. B / Vol. 24, No. 7 / July 2007
The fact that the shelf under the soliton oscillates at
the frequency of the soliton was also shown by Yang [12]
for the vector soliton solutions of the coupled NLS equa-
tions for a birefringent fiber by examining the eigenfunc-
tions for the equations governing a small perturbation
from the soliton state.
This research was supported by the Engineering and
Physical Sciences Research Council under grant EP/
C548612/1.
REFERENCES
1. G. Assanto, M. Peccianti, and C. Conti, “Nemations: optical
spatial solitons in nematic liquid crystals,” Opt. Photonics
News Feb. 2003, pp. 44–48.
2. C. Conti, M. Peccianti, and G. Assanto, “Route to
nonlocality and observation of accessible solitons,” Phys.
Rev. Lett. 91, 073901 (2003).
3. C. Conti, M. Peccianti, and G. Assanto, “Observation of
optical spatial solitons in a highly nonlocal medium,” Phys.
Rev. Lett. 92, 113902 (2004).
4. M. Peccianti, C. Conti, G. Assanto, A. de Luca, and C.
Umeton, “Routing of anisotropic spatial solitons and
modulational instability in liquid crystals,” Nature 432,
733–737 (2004).
5. E. A. Kuznetsov and A. M. Rubenchik, “Soliton
stabilization in plasmas and hydrodynamics,” Phys. Rep.
142, 103–165 (1986).
6. W. L. Kath and N. F. Smyth, “Soliton evolution and
radiation loss for the nonlinear Schrödinger equation,”
Phys. Rev. E 51, 1484–1492 (1995).
Minzoni et al.
7. C. García Reimbert, A. A. Minzoni, and N. F. Smyth,
“Spatial soliton evolution in nematic liquid crystals in the
nonlinear local regime,” J. Opt. Soc. Am. B 23, 294–301
(2006).
8. C. García Reimbert, A. A. Minzoni, N. F. Smyth, and A. L.
Worthy, “Large-amplitude nematicon propagation in a
liquid crystal with local response,” J. Opt. Soc. Am. B 23,
2551–2558 (2006).
9. P. D. Rasmussen, O. Bang, and W. Królikowski, “Theory of
nonlocal soliton interaction in nematic liquid crystals,”
Phys. Rev. E 72, 066611 (2005).
10. A. I. Yakimenko, V. M. Lashkin, and O. O. Prikhodko,
“Dynamics of two-dimensional coherent structures in
nonlocal nonlinear media,” Phys. Rev. E 73, 066605 (2006).
11. M. Abramowitz and I. A. Stegun, Handbook of
Mathematical Functions with Formulas, Graphs, and
Mathematical Tables (Dover, 1972).
12. J. Yang, “Vector solitons and their internal oscillations in
birefringent nonlinear optical fibers,” Stud. Appl. Math. 98,
61–97 (1997).
13. B. Fornberg and G. B. Whitham, “A numerical and
theoretical study of certain nonlinear wave phenomena,”
Philos. Trans. R. Soc. London, Ser. A 289, 373–403 (1978).
14. W. Królikowski, O. Bang, N. I. Nikolov, D. Neshev, J.
Whyller, J. J. Rasmussen, and D. Edmundson, “Modulation
instability, solitons and beam propagation in spatially
nonlocal nonlinear media,” J. Opt. B: Quantum
Semiclassical Opt. 6, S288–S294 (2004).
15. Y. Huang, Q. Guo, and J. Cao, “Optical beams in lossy
non-local Kerr media,” Opt. Commun. 261, 175–180 (2006).
16. M. Peccianti, A. De Rossi, G. Assanto, A. De Luca, C.
Umeton, and I. C. Khoo, “Electrically assisted self-
confinement and waveguiding in planar nematic liquid
crystal cells,” Appl. Phys. Lett. 77, 7–9 (2000).